Design And Implementation Of News Title Generation System Of COVID-19 Based On Text Similarity

At the end of 2019,a sudden outbreak of COVID-19 swept the world.The Chinese government has taken various effective measures in a timely manner."When one side is in trouble,all sides support",and the national anti epidemic has achieved remarkable results.However,with the development trend of the epidemic and the trend of normalization,various reports and news about COVID-19 are pouring into the Internet.How to reorganize the news title according to the actual news report content and certain rules and needs,so that the news title can clearly,objectively and truly display the news content has become a research topic with practical application value.Automatic text generation technology has been widely used in search result preview,news summary generation and other scenarios.At present,the Title Automatic Generation Based on text automatic generation technology has more restrictions in format and word number,higher requirements for the coverage of news content,and less existing research results and applications.This thesis mainly studies the technology of automatic generation of news headlines based on text similarity and applies them to COVID-19 related news reports in the post epidemic era and has designed a prototype system of COVID-19 news headline automatic generation based on text similarity.The overall architecture of the system implemented in this thesis mainly includes data processing module,algorithm model module and human-computer interaction interface.The Text Rank-Word2 Vec algorithm module applied in this research is the core part of the system.Based on the Word2 Vec model,the sentence similarities are calculated by the algorithm according to the Text Rank algorithm process in the text,in order to mine the semantic similarity of each sentence in the text from the word adjacency,so as to extract high-quality news headlines with high coverage and good readability.Through simulation experiment and system test,it is verified that the news headlines generated by Text Rank-Word2 Vec algorithm have certain advantages in manual test and ROUGE index.The research and application of this thesis can effectively improve the efficiency of network users' understanding of COVID-19's real-time transmission status,expand the scope of COVID-19 related news and information dissemination,and avoid inefficient,false or duplicated epidemic news.